Runway foreign object detection using RGB

This paper presents an improved algorithm for foreign object debris (FOD) detection on the runway with several innovative techniques. The detection scheme incorporates four steps of geometric adjustment, background subtraction, clutter suppression and camouflage elimination. After geometric adjustment, the background model is built for each pixel with a set of RGB colour values taken in the past at the same location or in the neighborhood in the step of background subtraction. The background model samples are substituted randomly with an unfixed update period. Furthermore, the steps of clutter suppression and camouflage elimination are added to modify the segmentation map after background subtraction in order to increase the detection probability and decrease the false alarm rate. The overall algorithm is applied to the test data and real data on the runway. The results show that the RGB-based algorithm performs better than the classical gray-based techniques.